LuckyLukke/negotio_GRPO

该数据集包含多轮对话的信息，每个对话包括对话的起始者(starting_agent)、游戏类型(game)、对话内容(trajectory)、角色(role)等。此外，数据集还提供了模型代理(model_agent_1和model_agent_2)的标识、评估(evaluation)信息、对话中的词汇索引(token)、掩码(mask)以及收益(payoff)。数据集分为训练集和测试集，支持NLP任务如对话生成和模型评估。

This dataset contains information on multi-turn dialogues, including details such as the starter of the dialogue (starting_agent), type of game (game), content of the dialogue (trajectory), roles (role), etc. Additionally, the dataset provides identifiers for model agents (model_agent_1 and model_agent_2), evaluation information (evaluation), token indices in the dialogue (token), masks (mask), and payoffs. The dataset is split into training and test sets and is suitable for NLP tasks such as dialogue generation and model evaluation.